Utilization of Deep Reinforcement Learning for Discrete Resource Allocation Problem in Project Management – a Simulation Experiment

Abstract

This paper tests the applicability of deep reinforcement learning (DRL) algorithms to simulated problems of constrained discrete and online resource allocation in project management. DRL is an extensively researched method in various domains, although no similar case study was found when writing this paper. The hypothesis was that a carefully tuned RL agent could outperform an optimisation-based solution. The RL agents: VPG, AC, and PPO, were compared against a classic constrained optimisation algorithm in trials: "easy"/"moderate"/"hard" (70/50/30% average project success rate). Each trial consisted of 500 independent, stochastic simulations. The significance of the differences was checked using a Welch ANOVA on significance level alpha = 0.01, followed by post hoc comparisons for false-discovery control. The experiment revealed that the PPO agent performed significantly better in moderate and hard simulations than the optimisation approach and other RL methods.